| Summary: | 碩士 === 朝陽科技大學 === 資訊管理系碩士班 === 95 === In the research of music databases, a repeating pattern is an important
feature of music objects, which commonly used in analyzing the repeated part of
music data and looking for theme. Most of repeating patterns are key melodies
or easy to familiarize and remember for people. Non-trivial repeating patterns
exclude those repeating patterns included in other longer patterns such that they
can reduce the redundancy and speedup music search. So far, existing
approaches discover a repeating pattern in such a way that the sequence of notes
in a music object appears more than once in exactly matching. If we allow the
similar sequences with partial different notes to be a fault tolerant repeating
pattern, it can reduce the number of repeating patterns and can construct more
efficient music indexes. Furthermore, the number of fault tolerant non-trivial
repeating patterns can still be a large amount such that some researches claimed
that the longest repeating patterns that can be characterized as an important
feature of melody strings. Therefore, if we can discover the longest fault tolerant
repeating pattern in a music object, it can be more suitable to represent the
theme of the music. In the existing approaches, they are time consuming for
finding fault tolerance non-trivial repeating patterns in a music objects and none
of them discusses the discovering of the longest fault tolerance repeating
patterns. In this paper, we aim to develop efficient approaches to discover fault
tolerance non-trivial repeating patterns in a music object as well as to find the
longest repeating pattern. We expect that the applications of our approach can be
extended such as applying to the searching for significant repeating patterns on
the DNA or protein gene sequences.
|
|---|
